Ependymoma (EPN) is a brain tumor commonly presenting in childhood that remains fatal in most children. Intra-tumoral cellular heterogeneity in bulk-tumor samples significantly confounds our understanding of EPN biology, impeding development of effective therapy. We, therefore, use single-cell RNA sequencing, histology, and deconvolution to catalog cellular heterogeneity of the major childhood EPN subgroups. Analysis of PFA subgroup EPN reveals evidence of an undifferentiated progenitor subpopulation that either differentiates into subpopulations with ependymal cell characteristics or transitions into a mesenchymal subpopulation. Histological analysis reveals that progenitor and mesenchymal subpopulations co-localize in peri-necrotic zones. In conflict with current classification paradigms, relative PFA subpopulation proportions are shown to determine bulk-tumor-assigned subgroups. We provide an interactive online resource that facilitates exploration of the EPN single-cell dataset. This atlas of EPN cellular heterogeneity increases understanding of EPN biology.

Pediatric Adamantinomatous Craniopharyngiomas (ACPs) are histologically benign brain tumors that often follow an aggressive clinical course. Their suprasellar location leaves them in close proximity to critical neurological and vascular structures and often results in significant neuroendocrine morbidity. Current treatment paradigms, involving surgical resection and radiotherapy, confer significant morbidity to patients and there is an obvious need to discover effective and safe alternative treatments. Recent years have witnessed significant efforts to fully detail the genomic, transcriptomic and proteomic make-up of these tumors, in an attempt to identify potential therapeutic targets. These studies have resulted in ever mounting evidence that inflammatory processes and the immune response play a critical role in the pathogenesis of both the solid and cystic portion of ACPs. Several inflammatory and immune markers have been identified in both the cyst fluid and solid tumor tissue of ACP. Due to the existence of effective agents that target them, IL-6 and immune checkpoint inhibitors seem to present the most likely immediate candidates for clinical trials of targeted immune-related therapy in ACP. If effective, such agents may result in a paradigm shift in treatment that ultimately reduces morbidity and results in better outcomes for our patients.

Deep learning (DL) is a widely applied mathematical modeling technique. Classically, DL models utilize large volumes of training data, which are not available in many healthcare contexts. For patients with brain tumors, non-invasive diagnosis would represent a substantial clinical advance, potentially sparing patients from the risks associated with surgical intervention on the brain. Such an approach will depend upon highly accurate models built using the limited datasets that are available. Herein, we present a novel genetic algorithm (GA) that identifies optimal architecture parameters using feature embeddings from state-of-the-art image classification networks to identify the pediatric brain tumor, adamantinomatous craniopharyngioma (ACP). We optimized classification models for preoperative Computed Tomography (CT), Magnetic Resonance Imaging (MRI), and combined CT and MRI datasets with demonstrated test accuracies of 85.3%, 83.3%, and 87.8%, respectively. Notably, our GA improved baseline model performance by up to 38%. This work advances DL and its applications within healthcare by identifying optimized networks in small-scale data contexts. The proposed system is easily implementable and scalable for non-invasive computer-aided diagnosis, even for uncommon diseases.

Background: Cellular senescence can have positive and negative effects on the body, including aiding in damage repair and facilitating tumor growth. Adamantinomatous craniopharyngioma (ACP), the most common pediatric sellar/suprasellar brain tumor, poses significant treatment challenges. Recent studies suggest that senescent cells in ACP tumors may contribute to tumor growth and invasion by releasing a senesecence-associated secretory phenotype. However, a detailed analysis of these characteristics has yet to be completed.Methods: We analyzed primary tissue samples from ACP patients using single-cell, single-nuclei, and spatial RNA sequencing. We performed various analyses, including gene expression clustering, inferred senescence cells from gene expression, and conducted cytokine signaling inference. We utilized LASSO to select essential gene expression pathways associated with senescence. Finally, we validated our findings through immunostaining.Results: We observed significant diversity in gene expression and tissue structure. Key factors such as NFKB, RELA, and SP1 are essential in regulating gene expression, while senescence markers are present throughout the tissue. SPP1 is the most significant cytokine signaling network among ACP cells, while the Wnt signaling pathway predominantly occurs between epithelial and glial cells. Our research has identified links between senescence-associated features and pathways, such as PI3K/Akt/mTOR, MYC, FZD, and Hedgehog, with increased P53 expression associated with senescence in these cells.Conclusions: A complex interplay between cellular senescence, cytokine signaling, and gene expression pathways underlies ACP development. Further research is crucial to understand how these elements interact to create novel therapeutic approaches for patients with ACP.

Diagnosis of adamantinomatous craniopharyngioma (ACP) is predominantly determined through invasive pathological examination of a neurosurgical biopsy specimen. Clinical experts can distinguish ACP from Magnetic Resonance Imaging (MRI) with an accuracy of 86%, and 9% of ACP cases are diagnosed this way. Classification using deep learning (DL) provides a solution to support a non-invasive diagnosis of ACP through neuroimaging, but it is still limited in implementation, a major reason being the lack of predictive uncertainty representation. We trained and tested a DL classifier on preoperative MRI from 86 suprasellar tumor patients across multiple institutions. We then applied a Bayesian DL approach to calibrate our previously published ACP classifier, extending beyond point-estimate predictions to predictive distributions. Our original classifier outperforms random forest and XGBoost models in classifying ACP. The calibrated classifier underperformed our previously published results, indicating that the original model was overfit. Mean values of the predictive distributions were not informative regarding model uncertainty. However, the variance of predictive distributions was indicative of predictive uncertainty. We developed an algorithm to incorporate predicted values and the associated uncertainty to create a classification abstention mechanism. Our model accuracy improved from 80.8% to 95.5%, with a 34.2% abstention rate. We demonstrated that calibration of DL models can be used to estimate predictive uncertainty, which may enable clinical translation of artificial intelligence to support non-invasive diagnosis of brain tumors in the future.